Congestion avoidance and control
SIGCOMM '88 Symposium proceedings on Communications architectures and protocols
Simulation-based comparisons of Tahoe, Reno and SACK TCP
ACM SIGCOMM Computer Communication Review
Improving the start-up behavior of a congestion control scheme for TCP
Conference proceedings on Applications, technologies, architectures, and protocols for computer communications
The performance of TCP/IP for networks with high bandwidth-delay products and random loss
IEEE/ACM Transactions on Networking (TON)
Comparative performance analysis of versions of TCP in a local network with a lossy link
IEEE/ACM Transactions on Networking (TON)
On the generation and use of TCP acknowledgments
ACM SIGCOMM Computer Communication Review
On TCP performance in a heterogeneous network: a survey
IEEE Communications Magazine
Resource Allocation with Persistent and Transient Flows
NETWORKING '02 Proceedings of the Second International IFIP-TC6 Networking Conference on Networking Technologies, Services, and Protocols; Performance of Computer and Communication Networks; and Mobile and Wireless Communications
TCP self-clocking and bandwidth sharing
Computer Networks: The International Journal of Computer and Telecommunications Networking
Revisiting the Performance of Short TCP Transfers
NETWORKING '09 Proceedings of the 8th International IFIP-TC 6 Networking Conference
A dynamic threshold and subsection control TCP slow-start algorithm
ICESS'05 Proceedings of the Second international conference on Embedded Software and Systems
FavorQueue: A parameterless active queue management to improve TCP traffic performance
Computer Networks: The International Journal of Computer and Telecommunications Networking
Hi-index | 0.00 |
Many works have studied the negative impact of slow start on the performance of short transfers. Some works propose to accelerate the window increase during this phase in order to improve the performance especially on satellite links. Others propose to set the slow start threshold at the beginning of the connection to a more accurate value in order to avoid losses. But, these works didn't account for the impact of network buffers on the performance. It is known that small buffers along with a fast window increase leads to an early buffer overflow and an underestimation of the network capacity. This may change completely the performance predicted by these modifications. In this paper, we present a general analysis of this first phase as a function of all the possible parameters. We show that, as claimed, the previous works improve the performance on paths with large buffers. However, on paths with small buffers, completely different results could be obtained.